JP7298687B2 - Object recognition device and object recognition method - Google Patents

Description

The present invention relates to technology for recognizing a three-dimensional object by template matching.

Template matching is one of methods for recognizing (detecting) an object from an image. Template matching is a method of detecting an object contained in an input image by preparing a model (template) of an object to be recognized in advance and evaluating the matching degree of image features between the input image and the model. be. Object recognition by template matching is used in a wide variety of fields, such as inspection and picking in FA (Factory Automation), robot vision, and monitoring cameras.

In recent years, attention has been focused on techniques that apply template matching to recognition of the three-dimensional position and orientation of an object. Its basic principle is to prepare a large number of templates with different views by changing the viewpoint position for the target object, and select the one that best matches the view of the target object in the input image from those templates. Thus, the three-dimensional position and orientation of the target object with respect to the camera are specified. However, in this method, the resolution of recognition is proportional to the variation of the template, so if you try to increase the resolution of recognition, problems such as an increase in the load of template creation, an increase in the amount of template data, and an increase in the processing time for template matching will occur. become prominent.

As a countermeasure against such problems, Patent Document 1 discloses that the depth distance of a target object is measured by a depth sensor, and a template (two-dimensional grid for sampling feature values) is scaled (enlarged/reduced) according to the depth distance. ) is disclosed.

U.S. Pat. No. 9,659,217

According to the method of Japanese Patent Laid-Open No. 2002-200312, since templates for a plurality of views that differ only in depth distance can be shared, effects such as reduction in the load of template creation and reduction in the number of templates can be expected. However, when searching for template matching, a process of enlarging or reducing the template according to the depth distance of each pixel occurs, so there is a demerit that the processing speed is slowed down. In order to reduce the time required to enlarge or reduce the template, generate multiple scale templates according to the distance range where the target object can exist and the required resolution and store them in the work memory prior to the template matching process. Although it is technically possible, it is impractical because it requires a very large memory capacity.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a practical technique that enables high-speed detection of objects that can exist at various depths by template matching.

One aspect of the present invention is a three-dimensional data acquisition unit that acquires three-dimensional data composed of a plurality of points each having three-dimensional information, and parallel projection of each point of the three-dimensional data onto a certain projection plane. and a recognition processing unit for detecting a target object from the two-dimensional image by template matching.

The three-dimensional data is preferably data obtained by three-dimensional measurement. Any three-dimensional measurement method may be used, and may be an active measurement method or a passive measurement method. Template matching evaluates the degree of matching (similarity) of image features between a template (model) of a target object and a region of interest in a two-dimensional image, so that a partial image within the region of interest is the image of the target object. It is a method of judging whether or not there is. By using a plurality of templates with different views of the target object for template matching, it is possible to recognize the posture of the target object.

In the present invention, a two-dimensional image generated by parallel projection of three-dimensional data is used for template matching. In parallel projection, the target object is projected in the same size regardless of the distance from the projection plane to the target object. Therefore, in a two-dimensional image generated by parallel projection, the image of the target object always has the same size (regardless of its depth distance). Therefore, since matching can be performed using only templates of a single size, high-speed processing is possible compared to conventional methods (methods in which templates are scaled according to depth distances). In addition, the number of templates and the amount of data can be reduced, and the required amount of work memory can be reduced, so there is also the advantage of being excellent in practicality.

The recognition processing unit may use a template generated from an image obtained by parallel projection of the target object as the template of the target object. By also generating the template from the parallel projection image, the matching accuracy between the template and the target object image in the two-dimensional image is improved, so that the reliability of the object recognition processing can be enhanced.

The projection plane may be set arbitrarily, but it is preferable to set the projection plane so that the projection points of the points forming the three-dimensional data are distributed as wide as possible on the projection plane. For example, when the three-dimensional data is data generated using an image captured by a camera, the parallel projection conversion unit sets the projection plane so as to be orthogonal to the optical axis of the camera. may

The parallel projection conversion unit converts depth information obtained from the three-dimensional information of the first point into the first pixel when the first point in the three-dimensional data is projected onto the first pixel in the two-dimensional image. pixels. When each point of the three-dimensional data has luminance information, the parallel projection conversion unit converts the first point of the three-dimensional data into the first pixel of the two-dimensional image. In some cases, the luminance information of the first point may be associated with the first pixel. When each point of the three-dimensional data has color information, the parallel projection conversion unit converts the first point of the three-dimensional data into the first pixel of the two-dimensional image. In some cases, the color information of the first point may be associated with the first pixel.

When there is no point to be projected onto the second pixel in the two-dimensional image, the parallel projection conversion unit converts the second pixel into may generate information associated with For example, the parallel projection conversion unit may obtain information associated with the second pixel by interpolating information associated with pixels around the second pixel. By increasing the amount of information in the two-dimensional image through such processing, an improvement in the accuracy of template matching can be expected.

The three-dimensional data is data generated using an image captured by a camera, and the parallel projection conversion unit projects a plurality of points in the three-dimensional data onto the same position on the projection plane. Alternatively, a point closest to the camera among the plurality of points may be used to generate the two-dimensional image. This processing generates a parallel projection image that considers overlapping (hidden) objects when viewed from the projection plane side (that is, only the points visible from the camera are mapped to the two-dimensional image). , object recognition processing by template matching can be performed with high accuracy.

The present invention may be regarded as an object recognition device having at least a part of the above-described means or configuration, or as an image processing device that performs the above-described parallel projection transformation. In addition, the present invention may be regarded as an object recognition method, an image processing method, a template matching method, an object recognition device control method, etc., including at least a part of the above processing, or a program or program for realizing such a method. It can also be regarded as a recording medium in which the program is non-temporarily recorded. It should be noted that each of the means and processes described above can be combined with each other as much as possible to constitute the present invention.

According to the present invention, it is possible to provide a practical technique that enables high-speed detection of objects that can exist at various depth distances by template matching.

FIG. 1 is a diagram schematically showing processing by an object recognition device. FIG. 2 is a diagram schematically showing the overall configuration of the object recognition device. FIG. 3 is a block diagram showing the configuration of the image processing apparatus. FIG. 4 is a flowchart of template creation processing. FIG. 5 is a diagram showing a setting example of viewpoint positions. FIG. 6 is a diagram showing an example of parallel projection images in the template creation process. FIG. 7 is a flowchart of object recognition processing. FIG. 8 is a flowchart of parallel projection conversion in object recognition processing. FIG. 9 is a diagram showing a setting example of the camera coordinate system and the projection image coordinate system. FIG. 10 is a flowchart of projection point complementing processing.

<Application example>
FIG. 1 schematically shows processing by an object recognition device, which is one of application examples of the present invention. Reference numeral 10 in FIG. 1 indicates how three objects 102a, 102b, and 102c on the stage 101 are measured (photographed) by the camera 103 from obliquely above. The objects 102a, 102b, and 102c have the same shape (cylindrical shape) and the same size, but the depth distance from the camera 103 is farther in the order of the object 102a, the object 102b, and the object 102c.

Reference numeral 11 is an example of three-dimensional data generated based on the image captured by the camera 103 . The three-dimensional data 11 is data composed of a plurality of points each having three-dimensional information. The three-dimensional data 11 may have any format, for example, data in which each point has a three-dimensional coordinate value, or a depth value (depth distance information) for each point (each pixel) of a two-dimensional image. ) may be associated with the data. The three-dimensional coordinate values may be coordinate values in the camera coordinate system, coordinate values in the global coordinate system, or coordinate values in another coordinate system. The three-dimensional data 11 in FIG. 1 is an example of a depth image, and the depth values are represented by gradation for convenience (a point farther from the camera 103 is darker). In a general optical system, an object farther from the camera 103 is formed into a smaller image, so the sizes on the image become smaller in the order of the object 102a, the object 102b, and the object 102c.

In conventional template matching, in order to deal with objects of various sizes, a plurality of types of templates with different sizes are used, or template sizes are scaled according to depth values as in Patent Document 1. However, these conventional methods have the disadvantage of causing problems such as a decrease in processing speed and an increase in memory capacity, as described above.

Therefore, in the embodiment of the present invention, the three-dimensional data 11 is subjected to parallel projection transformation to generate a two-dimensional image 12, and this two-dimensional image 12 is used for template matching. Objects having the same actual size also have the same size on the two-dimensional image 12 by performing the parallel projection transformation. Therefore, all objects 102a, 102b, 102c contained in the two-dimensional image 12 can be detected by applying a template 13 of a single size. Reference numeral 14 indicates an example of recognition results.

According to the method of this embodiment, high-speed processing is possible compared to the conventional method. In addition, the number of templates and the amount of data can be reduced, and the required amount of work memory can be reduced, so there is also the advantage of being excellent in practicality. For convenience of explanation, FIG. 1 shows an example in which the objects 102a, 102b, and 102c have the same orientation. A template 13 may be prepared for each posture.

<Embodiment>
(Overall configuration of object recognition device)
An object recognition device according to an embodiment of the present invention will be described with reference to FIG.

The object recognition device 2 is installed in a production line that assembles and processes articles, and uses data captured from the sensor unit 20 to recognize the position and orientation of an object 27 loaded on a tray 26 by template matching ( 3D object recognition). Objects to be recognized (hereinafter also referred to as “target objects”) 27 are randomly stacked on the tray 26 .

The object recognition device 2 is roughly composed of a sensor unit 20 and an image processing device 21 . A wired or wireless connection is established between the sensor unit 20 and the image processing device 21 , and the output of the sensor unit 20 is taken into the image processing device 21 . The image processing device 21 is a device that performs various processes using data captured from the sensor unit 20 . The processing of the image processing device 21 may include, for example, distance measurement (distance measurement), three-dimensional shape recognition, object recognition, scene recognition, and the like. A recognition result of the object recognition device 2 is output to a PLC (programmable logic controller) 25, a display 22, or the like, for example. The recognition results are used, for example, for control of the picking robot 28, control of processing devices and printing devices, inspection and measurement of the target object 27, and the like.

(sensor unit)
The sensor unit 20 has at least a camera for taking an optical image of the target object 27 . Furthermore, the sensor unit 20 may include components (sensors, illumination devices, light projection devices, etc.) necessary for three-dimensional measurement of the target object 27 . For example, when the depth distance is measured by stereo matching (also called stereo vision, stereo camera method, etc.), the sensor unit 20 is provided with a plurality of cameras. In the case of active stereo, the sensor unit 20 is further provided with a light projection device for projecting pattern light onto the target object 27 . When three-dimensional measurement is performed by the space-encoded pattern projection method, the sensor unit 20 is provided with a light projection device for projecting pattern light and a camera. In addition, any method such as a photometric stereo method, a TOF (time of flight) method, a phase shift method, or the like may be used as long as the method can acquire the three-dimensional information of the target object 27 .

(Image processing device)
The image processing device 21 is configured by a computer including, for example, a CPU (processor), RAM (memory), nonvolatile storage device (hard disk, SSD, etc.), input device, output device, and the like. In this case, the CPU develops a program stored in the nonvolatile storage device in the RAM and executes the program, thereby realizing various configurations described later. However, the configuration of the image processing device 21 is not limited to this, and all or part of the configuration described later may be realized by a dedicated circuit such as FPGA or ASIC, or by cloud computing or distributed computing. You may

FIG. 3 is a block diagram showing the configuration of the image processing device 21. As shown in FIG. The image processing device 21 has the configuration of the template creation device 30 and the configuration of the object recognition processing device 31 . The template creation device 30 is a configuration for creating a template used in object recognition processing, and includes a three-dimensional CAD data acquisition unit 300, a parallel projection parameter setting unit 301, a viewpoint position setting unit 302, and a two-dimensional projection image creation unit 303. , a feature extraction unit 304 and a template creation unit 305 . The object recognition processing device 31 is configured to execute object recognition processing by template matching, and includes a three-dimensional data acquisition unit 310, a parallel projection parameter setting unit 311, a parallel projection conversion unit 312, a feature extraction unit 313, and a template storage unit. 314 , a template matching unit 315 , and a recognition result output unit 316 . In this embodiment, the feature extraction unit 313, the template storage unit 314, and the template matching unit 315 constitute the "recognition processing unit" of the present invention.

(Template creation process)
An example of template creation processing by the template creation device 30 will be described with reference to the flowchart of FIG.

In step S<b>400 , the three-dimensional CAD data acquisition unit 300 acquires three-dimensional CAD data of the target object 27 . The CAD data may be read from the internal storage device of the image processing apparatus 21, or may be obtained from an external CAD system, storage, or the like via a network. Note that three-dimensional shape data measured by a three-dimensional sensor or the like may be obtained instead of CAD data.

In step S401, the viewpoint position setting unit 302 sets a viewpoint position for creating a template. FIG. 5 shows an example of setting viewpoint positions. In this example, viewpoints (indicated by black circles) are set at 42 vertices of an octahedron including the target object 27 . Note that the number and arrangement of viewpoints may be appropriately set according to the required resolution, the shape of the target object 27, possible postures, and the like. The number and arrangement of viewpoints may be specified by the user, or may be automatically set by the viewpoint position setting unit 302 .

In step S402, the parallel projection parameter setting unit 301 sets parallel projection parameters used for template creation. Here, two parameters res _x and res _y are used as parallel projection parameters. (res _x , res _y ) is the size of one pixel of the projected image (unit: mm). Parallel projection parameters are also used in parallel projection transformation in object recognition processing, which will be described later. By aligning the values of the parallel projection parameters, the size of the target object 27 in the template matches the size of the target object 27 in the parallel projection image generated by the object recognition processing. This is because it eliminates the need for adjustment.

In step S403, the two-dimensional projection image creation unit 303 creates a two-dimensional projection image by parallel projection of the three-dimensional CAD data. FIG. 6 shows an example of a two-dimensional projection image. A two-dimensional projection image 60 corresponding to the viewpoint VP is created by parallel-projecting each point on the surface of the target object 27 onto a projection plane 62 passing through the viewpoint VP.

In step S404, the feature extraction unit 304 extracts image features of the target object 27 from the two-dimensional projection image 60 created in step S403. Image features include, for example, brightness, color, brightness gradient direction, quantization gradient direction, HoG (Histogram of Oriented Gradients), surface normal direction, HAAR-like, SIFT (Scale-Invariant Feature Transform), etc. can be done. The intensity gradient direction represents the direction (angle) of the intensity gradient in a local area centered on a feature point as a continuous value. Gradient directions are represented by discrete values (for example, 8 directions are stored as 1-byte information from 0 to 7). The feature extraction unit 304 may obtain image features for all points (pixels) of the two-dimensional projection image 60, or may obtain image features for some points sampled according to a predetermined rule. A point from which an image feature is obtained is called a feature point.

In step S405, the template creation unit 305 creates a template corresponding to the viewpoint VP based on the image features extracted in step S404. A template is, for example, a data set including coordinate values of each feature point and extracted image features.

The processing of steps S403 to S405 is performed for all viewpoints set in step S401 (step S406). When template creation for all viewpoints is completed, the template creation unit 305 stores template data in the template storage unit 314 of the object recognition processing device 31 (step S407). This completes the template creation process.

(Object recognition processing)
An example of object recognition processing by the object recognition processing device 31 will be described with reference to the flowchart of FIG.

In step S<b>700 , the three-dimensional data acquisition section 310 generates three-dimensional data within the field of view based on the image captured by the sensor unit 20 . In the present embodiment, stereo images are captured by two cameras while pattern light is projected from the projection device, and the depth distance is calculated based on the parallax between the images. Obtain 3D information.

In step S701, the parallel projection parameter setting unit 311 sets parallel projection parameters used for parallel projection conversion. Here, four parameters of res _x , res _y , c _x , and c _y are used as parallel projection parameters. (res _x , res _y ) is the size of one pixel of the projection image (unit: mm), and may be set to any value. For example, using the focal length (f _x , f _y ) of the camera of the sensor unit 20,
_resx = d/ _fx
res _y =d/f _y
may be d is a constant set according to the depth distance in which the target object 27 can exist. For example, the average value, minimum value, or maximum value of the depth distance from the sensor unit 20 to the target object 27 may be set as the constant d. As described above, it is preferable to use the same values for (res _x , res _y ) as those used when creating the template. (c _x , c _y ) are the center coordinates of the projection image.

In step S702, the parallel projection conversion unit 312 generates a two-dimensional projection image by parallel-projecting each point in the three-dimensional data (hereinafter referred to as "three-dimensional point") onto a predetermined projection plane.

Details of the parallel projection transformation will be described with reference to FIGS. 8 and 9. FIG. In step S800, the parallel projection conversion unit 312 calculates image coordinate values when a three-dimensional point is parallel projected. Assume that the camera coordinate system is (X, Y, Z) and the image coordinate system of the projection image is (x, y). In the example of FIG. 9, the origin O coincides with the center (principal point) of the lens of the camera of the sensor unit 20, the Z axis overlaps the optical axis, and the X axis and Y axis are the horizontal and vertical directions of the imaging element of the camera. A camera coordinate system is set so as to be parallel to . The image coordinate system is set so that the image center (c _x , c _y ) is on the Z axis of the camera coordinate system, and the x and y axes are parallel to the X and Y axes of the camera coordinate system, respectively. be done. The xy plane of the image coordinate system is the projection plane. That is, in this embodiment, the projection plane for parallel projection conversion is set so as to be orthogonal to the optical axis of the camera. When the coordinate system is set as shown in FIG. 9, the image coordinate values (x i , y _i ) after parallel projection transformation corresponding to the three-dimensional point (X _{i ,} Y _i , Z _i ) are
x _i =ROUND(X _i /res _x +c _x )
y _i =ROUND(Y _i /res _y +c _y )
Determined by ROUND is an operator for rounding to decimal places.

In step S801, the parallel projection transformation unit 312 checks whether a three-dimensional point projected onto the image coordinate values (x _i , y _i ) already exists. Specifically, it is checked whether the pixel (x _i , y _i ) of the projection image is already associated with 3D point information. If there is no associated three-dimensional point yet (NO in step S801), the parallel projection conversion unit 312 converts the information of the three-dimensional point (X _i , Y _i , Z _i ) to the pixel (x _i , y _i ). are associated (step S803). In this embodiment, the coordinate values (X _i , Y _i , Z _i ) of the three-dimensional point are associated with the pixel (x _i , y _{i )} . values), color information (eg, RGB values), luminance information, and the like. If the associated three-dimensional point _already exists (YES in step S801), the parallel projection transformation unit 312 compares the value of Z _i with the already associated Z value, If it is smaller (YES in step S802), the information associated with the pixel (x _i , y _i ) is overwritten with the information of the three-dimensional point (X _i , Y _i , Z _i ) (step S803). When a plurality of 3D points are projected onto the same position on the projection plane by such processing, the information of the 3D point closest to the camera among the plurality of 3D points is used to generate the projected image. It will happen. After the processing of steps S800 to S803 has been performed for all three-dimensional points, the process proceeds to step S703 in FIG. 7 (step S804).

In step S703, the feature extraction unit 313 extracts image features from the projection image. The image features extracted here are the same as the image features used to create the template. In step S704, the template matching unit 315 reads the template from the template storage unit 314, and detects the target object from the projection image by template matching processing using the template. At this time, the orientation of the target object can also be recognized by using templates of different viewpoints. In step S705, the recognition result output unit 316 outputs the recognition result. The object recognition processing ends here.

(Advantages of this embodiment)
In the configuration and processing described above, a two-dimensional image generated by parallel projection of three-dimensional data is used for template matching. In parallel projection, the target object is projected in the same size regardless of the distance from the projection plane to the target object. Therefore, in a two-dimensional image generated by parallel projection, the image of the target object always has the same size (regardless of its depth distance). Therefore, since matching can be performed using only templates of a single size, high-speed processing is possible compared to the conventional method. In addition, the number of templates and the amount of data can be reduced, and the required amount of work memory can be reduced, so there is also the advantage of being excellent in practicality.

Further, in this embodiment, since the template is also generated from the parallel projection image, the accuracy of matching between the template and the target object image in the image obtained by the parallel projection conversion is improved. Thereby, the reliability of object recognition processing can be improved.

In addition, in this embodiment, the projection plane is set so as to be orthogonal to the optical axis of the camera. Therefore, the calculation of the conversion from the camera coordinate system to the image coordinate system can be simplified, and the balanced projection conversion process can be speeded up. It is possible to increase the speed of object recognition processing by matching. Further, by setting the projection plane so as to be orthogonal to the optical axis of the camera, it is possible to suppress the distortion of the target object image after parallel projection conversion.

Also, when a plurality of 3D points are projected onto the same pixel, only the information of the 3D point closest to the camera is used. A parallel projection image that takes into consideration the object is generated, and object recognition processing by template matching can be performed with high accuracy.

<Others>
The above-described embodiment is merely an example of the configuration of the present invention. The present invention is not limited to the specific forms described above, and various modifications are possible within the technical scope of the present invention.

For example, the projection point complementing process shown in FIG. 10 may be performed after the parallel projection conversion process (step S702 in FIG. 7). Specifically, the parallel projection conversion unit 312 checks whether each pixel (x _i , y _i ) of the projection image generated in step S702 is associated with three-dimensional point information (step S100). , when the information of the 3D point is not associated (that is, when the projection point does not exist), the pixel (x _i , y _i ) is associated with the surrounding pixels (for example, 4 neighboring pixels, 8 neighboring pixels, etc.) Information for the pixel (x _i , y _i ) is generated based on the information stored (step S101). For example, nearest neighbor, bilinear, bicubic, etc. interpolation may generate information for pixel (x _i , y _i ). Then, the parallel projection transformation unit 312 associates the information generated in step S101 with the pixel (x _i , y _i ) (step S102). The processing of steps S100 to S102 is performed for all pixels of the projection image. Such processing increases the amount of information (the number of projection points) of the projected image, and thus can be expected to improve the accuracy of template matching.

Also, the setting of the projection plane is not limited to the example in FIG. For example, the projection plane may be arranged on the back side (image side) of the origin O of the camera coordinate system. Alternatively, the projection plane may be arranged so that it obliquely intersects the optical axis (Z-axis) (that is, the projection direction is non-parallel to the optical axis).

<Appendix>
(1) a three-dimensional data acquisition unit (310) for acquiring three-dimensional data composed of a plurality of points each having three-dimensional information;
a parallel projection transformation unit (312) that generates a two-dimensional image by parallel-projecting each point of the three-dimensional data onto a certain projection plane;
a recognition processing unit (313, 314, 315) that detects a target object from the two-dimensional image by template matching;
An object recognition device (2) characterized by comprising:

2: Object recognition device 20: Sensor unit 21: Image processing device 22: Display 27: Target object 30: Template creation device 31: Object recognition processing device

Claims (13)

a three-dimensional data acquisition unit that acquires three-dimensional data composed of a plurality of points each having three-dimensional information;
a parallel projection conversion unit that generates a two-dimensional image by parallel-projecting each point of the three-dimensional data onto a certain projection plane;
a recognition processing unit that detects a target object from the two-dimensional image by template matching ;
The three-dimensional data is data generated using an image captured by a camera,
The object recognition apparatus , wherein the parallel projection conversion unit sets the projection plane so as to be orthogonal to the optical axis of the camera . a three-dimensional data acquisition unit that acquires three-dimensional data composed of a plurality of points each having three-dimensional information;
a parallel projection conversion unit that generates a two-dimensional image by parallel-projecting each point of the three-dimensional data onto a certain projection plane;
a recognition processing unit that detects a target object from the two-dimensional image by template matching;
Each point of the three-dimensional data has luminance information,
The parallel projection conversion unit associates luminance information of the first point with the first pixel when the first point in the three-dimensional data is projected onto the first pixel in the two-dimensional image.
An object recognition device characterized by: a three-dimensional data acquisition unit that acquires three-dimensional data composed of a plurality of points each having three-dimensional information;
a parallel projection conversion unit that generates a two-dimensional image by parallel-projecting each point of the three-dimensional data onto a certain projection plane;
a recognition processing unit that detects a target object from the two-dimensional image by template matching;
has
Each point of the three-dimensional data has color information,
The parallel projection conversion unit associates color information of the first point with the first pixel when the first point in the three-dimensional data is projected onto the first pixel in the two-dimensional image. An object recognition device characterized by: a three-dimensional data acquisition unit that acquires three-dimensional data composed of a plurality of points each having three-dimensional information;
a parallel projection conversion unit that generates a two-dimensional image by parallel-projecting each point of the three-dimensional data onto a certain projection plane;
a recognition processing unit that detects a target object from the two-dimensional image by template matching;
The three-dimensional data is data generated using an image captured by a camera,
When a plurality of points in the three-dimensional data are projected onto the same position on the projection plane, the parallel projection conversion section converts a point closest to the camera from among the plurality of points to generate the two-dimensional image. used for
An object recognition device characterized by: The parallel projection conversion unit converts depth information obtained from the three-dimensional information of the first point into the first pixel when the first point in the three-dimensional data is projected onto the first pixel in the two-dimensional image. associated with pixels of
The object recognition device according to any one of claims 1 to 4, characterized in that: When there is no point to be projected onto the second pixel in the two-dimensional image, the parallel projection conversion unit converts the second pixel into generate information associated with
6. The object recognition device according to claim 2, 3, or 5, characterized in that: The parallel projection conversion unit obtains information associated with the second pixel by interpolating information associated with pixels surrounding the second pixel.
7. The object recognition device according to claim 6, characterized by: The object recognition apparatus according to any one of claims 1 to 7, wherein the recognition processing unit uses, as a template of the target object, a template generated from an image obtained by parallel projection of the target object. . obtaining three-dimensional data consisting of a plurality of points each having three-dimensional information;
generating a two-dimensional image by parallel projection of each point of the three-dimensional data onto a projection plane;
detecting a target object from the two-dimensional image by template matching;
has
The three-dimensional data is data generated using an image captured by a camera,
In the parallel projection, the object recognition method is characterized in that the projection plane is set so as to be orthogonal to the optical axis of the camera . obtaining three-dimensional data consisting of a plurality of points each having three-dimensional information;
generating a two-dimensional image by parallel projection of each point of the three-dimensional data onto a projection plane;
detecting a target object from the two-dimensional image by template matching;
has
Each point of the three-dimensional data has luminance information,
The parallel projection associates luminance information of the first point with the first pixel when the first point in the three-dimensional data is projected onto the first pixel in the two-dimensional image.
An object recognition method characterized by: obtaining three-dimensional data consisting of a plurality of points each having three-dimensional information;
generating a two-dimensional image by parallel projection of each point of the three-dimensional data onto a projection plane;
detecting a target object from the two-dimensional image by template matching;
has
Each point of the three-dimensional data has color information,
The parallel projection associates color information of the first point with the first pixel when the first point in the three-dimensional data is projected onto the first pixel in the two-dimensional image.
An object recognition method characterized by: obtaining three-dimensional data consisting of a plurality of points each having three-dimensional information;
generating a two-dimensional image by parallel projection of each point of the three-dimensional data onto a projection plane;
detecting a target object from the two-dimensional image by template matching;
has
The three-dimensional data is data generated using an image captured by a camera,
In the parallel projection, when a plurality of points in the three-dimensional data are projected onto the same position on the projection plane, the point closest to the camera among the plurality of points is used to generate the two-dimensional image.
An object recognition method characterized by: A program for causing a computer to execute each step of the object recognition method according to any one of claims 9 to 12 .

Images